DiceDB MCP

# DiceDB MCP A [Model Context Protocol (MCP)](https://modelcontextprotocol.io) server implementation for DiceDB to enable interactions between AI applications (hosts/clients) and DiceDB database servers. This implementation uses the [DiceDB Go SDK](https://github.com/DiceDB/dicedb-go) to communicate with DiceDB. Check out the [demo video](./demo.mov) to see it in action! ## Features - PING DiceDB to check connectivity. - ECHO a message through DiceDB. - GET a value from DiceDB by key. - SET a key-value pair in DiceDB. - DEL one or more keys from DiceDB. - INCR the integer value of a key by one. - DECR the integer value of a key by one. ## Installation ### Download Binary You can [download](https://github.com/pottekkat/dicedb-mcp/releases) and use the appropriate binary for your operating system and processor archetecture from the "Releases" page. ### Install via Go Prerequisites: - Go 1.24 or higher ```bash go install github.com/pottekkat/dicedb-mcp@latest ``` Get the path to the `dicedb-mcp` binary: ```bash which dicedb-mcp ``` ### Build from Source See [Development](#development) section below. ## Usage ### With MCP Hosts/Clients Add this to your `claude_desktop_config.json` for Claude Desktop or `mcp.json` for Cursor: ```json { "mcpServers": { "dicedb-mcp": { "command": "path/to/dicedb-mcp" } } } ``` ### With OpenAI Agents SDK The example below shows how to use the `dicedb-mcp` server with the [OpenAI Agents SDK](https://openai.github.io/openai-agents-python/): ```python from agents import Agent, Runner, trace from agents.mcp import MCPServer, MCPServerStdio from dotenv import load_dotenv import os import openai import asyncio load_dotenv() async def run(mcp_server: MCPServer, prompt: str, server_url: str): agent = Agent(name="DiceDB MCP", instructions=f"""You can interact with a DiceDB database running at {server_url}, use this for url.""", mcp_servers=[mcp_server],) result = await Runner.run(starting_agent=agent, input=prompt) print(result.final_output) async def main(): openai.api_key = os.getenv("OPENAI_API_KEY") prompt = "Can you change the value of the 'name' key to 'Rachel Green'?" server_url = "localhost:7379" async with MCPServerStdio( cache_tools_list=True, params={"command": "path/to/dicedb-mcp", "args": [""]}, ) as server: with trace(workflow_name="DiceDB MCP"): await run(server, prompt, server_url) if __name__ == "__main__": asyncio.run(main()) ``` ## Available Tools ### ping Pings a DiceDB server to check connectivity. ### echo Echoes a message through the DiceDB server. ### get Retrieves a value from DiceDB by key. ### set Sets a key-value pair in DiceDB. ### del Deletes one or more keys from DiceDB. ### incr Increments the integer value of a key by one. ### decr Decrements the integer value of a key by one. ## Development Fork and clone the repository: ```bash git clone https://github.com/username/dicedb-mcp.git ``` Change into the directory: ```bash cd dicedb-mcp ``` Install dependencies: ```bash make deps ``` Build the project: ```bash make build ``` Update your MCP servers configuration to point to the local build: ```json { "mcpServers": { "dicedb-mcp": { "command": "/path/to/dicedb-mcp/dist/dicedb-mcp" } } } ``` ## License [MIT License](LICENSE)